Article pubs.acs.org/jmc
Development of [18F]-Labeled Pyrazolo[4,3‑e]‑1,2,4triazolo[1,5‑c]pyrimidine (SCH442416) Analogs for the Imaging of Cerebral Adenosine A2A Receptors with Positron Emission Tomography Shivashankar Khanapur,† Soumen Paul,† Anup Shah,‡ Suresh Vatakuti,§ Michel J. B. Koole,† Rolf Zijlma,† Rudi A. J. O. Dierckx,† Gert Luurtsema,† Prabha Garg,‡ Aren van Waarde,† and Philip H. Elsinga*,† †
Department of Nuclear Medicine and Molecular Imaging, University Medical Center Groningen, University of Groningen, Groningen, The Netherlands ‡ Computer Centre, National Institute of Pharmaceutical Education and Research, SAS Nagar, Mohali, India § Department of Pharmacokinetics, Toxicology and Targeting, University of Groningen, Groningen, The Netherlands
ABSTRACT: Cerebral adenosine A2A receptors (A2ARs) are attractive therapeutic targets for the treatment of neurodegenerative and psychiatric disorders. We developed high affinity and selective compound 8 (SCH442416) analogs as in vivo probes for A2ARs using PET. We observed the A2AR-mediated accumulation of [18F]fluoropropyl ([18F]-10b) and [18F]fluoroethyl ([18F]10a) derivatives of 8 in the brain. The striatum was clearly visualized in PET and in vitro autoradiography images of control animals and was no longer visible after pretreatment with the A2AR subtype-selective antagonist KW6002. In vitro and in vivo metabolite analyses indicated the presence of hydrophilic (radio)metabolite(s), which are not expected to cross the blood-brainbarrier. [18F]-10b and [18F]-10a showed comparable striatum-to- cerebellum ratios (4.6 at 25 and 37 min post injection, respectively) and reversible binding in rat brains. We concluded that these compounds performed equally well, but their kinetics were slightly different. These molecules are potential tools for mapping cerebral A2ARs with PET.
■
(cAMP).4 The subtypes differ in size (A1, A2A, A2B, and A3 consist of 326, 409, 328, and 318 amino acids, respectively) and exhibit unique tissue distributions.5 Over the last 30 years, the most extensively studied AR subtypes are the biochemically and pharmacologically wellcharacterized high-affinity A1Rs and A2ARs. Adenosine activates these receptors at nanomolar concentrations.2 A1Rs are widely distributed in the human brain, with the highest densities being found in the hippocampus, cerebral cortex, thalamic nuclei, and dorsal horn of spinal cord, whereas A2ARs are highly expressed in the dopamine-rich regions of the brain, and the highest levels of A2AR expression occur in the striatum (caudate-putamen, nucleus accumbens, and olfactory tubercle), globus pallidus and substantia nigra.6−9 Lower levels of A2ARs occur in the
INTRODUCTION Adenosine, an endogenous signaling substance, is a purine ribonucleoside that is composed of adenine (purine base) and ribose (sugar molecule).1 It functions as a cytoprotectant and neuromodulator in response to organ and tissue stresses under both physiological and pathological conditions.2 In the brain, adenosine plays an important role in the regulation of both neuronal and glial cell functions. Furthermore, it counteracts glutamate excitotoxicity and cytokine-induced apoptosis.3 Its actions are mediated through the activation of four subtypes of G-protein coupled adenosine receptors (ARs), namely A1, A2A, A2B, and A3.2 Adenosine A1 receptors (A1Rs) and adenosine A3 receptors (A3Rs) are G-protein-coupled binding sites for adenosine that inhibit adenylyl cyclase, whereas adenosine A2A receptors (A2ARs) and adenosine A2B receptors (A2BRs) stimulate adenylyl cyclase via GS proteins and promote the formation of the second messenger cyclic adenosine monophosphate © 2014 American Chemical Society
Received: May 23, 2014 Published: July 25, 2014 6765
dx.doi.org/10.1021/jm500700y | J. Med. Chem. 2014, 57, 6765−6780
Journal of Medicinal Chemistry
Article
Figure 1. Structures of some selected AR agonists, an antagonist KW6002, and A2AR PET imaging agents.
PD24 or AD.25 Along with D2Rs, A2AR antagonists attenuate the overactivity of the indirect dopamine pathway observed during PD, restore balance between the direct and indirect output pathways, and suppress the neurodegenerative process by modulating the activity of cortico-striato-pallido-thalamocortical (CSPTC) pathways.20,26,27 Positron emission tomography (PET) can noninvasively assess the functional status of CSPTC pathways in PD and related disorders.28 High-affinity antagonistic radioligands that are selective for the A2ARs can be used to assess changes in A2AR density during disease progression and to monitor the effects of therapy on these changes.2 Moreover, they can be employed to assess the occupancy of the receptor population by therapeutic drugs in the human brain, which will allow the correlation of receptor occupancy with dose, drug/tracer plasma levels, and therapeutic effects.2,18,29,30 We focused on A2AR antagonist core structures as PET ligands (instead of A2AR agonist structures) because agonist PET tracers may only bind to the high-affinity state of the receptors, resulting in a poor signal-to-noise ratio.31 Moreover, the in vivo vulnerability to competition by endogenous adenosine may hamper the quantification of the total number of binding sites.31 The design and development of novel A2AR antagonist PET ligands is a key research topic because current xanthine-based tracers suffer from several disadvantages, including high nonspecific binding, low signal-to-noise ratios, and therefore, target sites in the brain are barely visible. Furthermore, the use of xanthine-based tracers suffer from a photoisomerization problem, low selectivity toward A2AR.2,29,32−38 On the basis of these considerations, nonxanthine compounds were developed and evaluated in many preclinical and clinical studies for the assessment of cerebral A2ARs.18
hippocampus, cerebral cortex, amygdala, cerebellum, brainstem, and hypothalamus.10−13 The activation of AlRs (e.g., by A1R agonists) increases sleep, inhibits seizures, reduces anxiety, and promotes neuroprotection. On the other hand, A1R antagonists are anxiolytics, are beneficial in the treatment of cognitive disorders, cardiac arrhythmia, asthma, and other respiratory disorders, and are therapeutic drugs for kidney protection.14 A2AR agonists are implicated in tissue repair, which involves a series of coordinated and overlapping phases such as inflammation, wound healing, angiogenesis, and tissue reorganization.15−17 The vasodilating effect of A2AR agonists (adenosine and regadenoson) has been fully validated (see Figure 1). Adenosine (Figure 1, compound 1) is used for the treatment of paroxysmal supraventricular tachycardia,3 whereas regadenoson (Figure 1, compound 2) is currently marketed for use as a pharmacologic stress agent in (radionuclide) myocardial perfusion imaging.18 A2ARs mediate potential neuroprotective and neurotoxic effects in addition to modulating dopaminergic neurotransmission in the basal ganglia through the antagonistic interactions between A2ARs and dopamine D2 receptors (D2Rs).19−21 However, their role in neurodegenerative diseases such as Parkinson’s disease (PD) and Alzheimer’s disease (AD) is highly controversial.21 Nevertheless, based on preclinical studies, A2AR antagonists have potential benefits in the treatment of neurodegenerative and psychiatric disorders such as PD, AD, neuroinflammation, ischemia, spinal cord injury, drug addiction, and other conditions.18,20−23 Moreover, recent epidemiological studies have established that the regular consumption of caffeine (a xanthine derivative and AR antagonist) is associated with a lower risk for developing 6766
dx.doi.org/10.1021/jm500700y | J. Med. Chem. 2014, 57, 6765−6780
Journal of Medicinal Chemistry
Article
A high degree of selectivity and appropriate combinations of lipophilicity, molecular weight, and affinity are important in the development of ideal in vivo A2AR PET brain tracers.39 For a compound to cross the blood-brain-barrier (BBB), a relatively small molecular weight (400 to 500 Da) and moderate lipophilicity (approximate range of log P is 2 to 3.5) are optimal.39−41 High lipophilicity causes unacceptable binding to plasma proteins and lipid bilayers, resulting in high levels of nonspecific binding in the brain.39 Low lipophilicity decreases the penetration of PET agents across the BBB. In addition, the tracer’s affinity must balance the opposing goals of tight binding and significant washout from the brain. Furthermore, the easy and rapid (within 3 half-lives) incorporation of radionuclides into the appropriate precursor molecules is necessary. Finally, the formation of lipophilic radioactive metabolites should be negligible because the presence of resulting radiometabolites in the target tissue would impede the quantification of the PET data using kinetic models.39,42 Rodent studies with nonxanthine tracer 7-(3-(4-[11C]methoxyphenyl)propyl)-2-(2-furyl)pyrazolo[4,3-e]-1,2,4triazolo[1,5-c]pyrimidine-5-amine [11C]-8 ([11C]SCH442416; Figure 1)43,44 and its 2-[18F]fluoroethyl derivative [18F]-10a ([18F]MRS5425 = [18F]FESCH in Scheme 3)47,48 suggest their potential value for in vivo mapping of A2ARs.43−48 However, high nonspecific binding and a lower striatum (target)-tocerebellum (nontarget) ratios (4.6 for [11C]-8 at 15 min; no target-to-nontarget data available for [18F]-10a) than [7methyl-11C](E)-8-(2,3-dimethyl-4-methoxystyryl)-1,3,7-trimethylxanthine [11C]-3 ([11C]KF21213; a xanthine-based tracer, 10.5 at 60 min, Figure 1)38 were associated with these tracers, and these compounds will require further evaluation in human subjects.44,48 Furthermore, [11C]-3 (a tight binding tracer suffers from several limitations such as photoisomerization, an impaired kinetic profile, low BBB penetration, and poor water solubility.2,44 All of the A2AR antagonist PET ligands that have been successfully evaluated in humans to date are [11C]-labeled. The radioisotope 18F has advantages of higher specific activity and a longer physical half-life (109.8 min vs 20.4 min) than [11C] ligands; these advantages allow the tracer to be distributed to remote imaging centers without cyclotron facilities and to achieve longer biodistribution and scanning times for a better assessment of the dissociation rate constant (Koff).48,49 On the basis of the above considerations, favorable docking results and because compound 8 demonstrated an appropriate lipophilicity for CNS imaging in addition to the highest A2AR affinity and subtype-selectivity, the present study utilized 8 as a lead compound to develop a novel radiofluorinated A2AR ligand. We designed and prepared an 7-(3-(4-(3-[18F]¯uoropropoxy)phenyl)propyl)-2-(furan-2-yl)-7H-pyrazolo[4,3-e][1,2,4]triazolo[1,5-c]pyrimidin-5-amine derivative [ 18 F]-10b ([18F]FPSCH in Scheme 3) and compared its pharmacokinetics and biodistribution in healthy rats with those of [18F]10a and [11C]-8 to optimize the length of the fluoroalkyl chain, which could affect both A2AR affinity and selectivity.
Table 1. Docking Analysis of A2A Antagonists Using GOLD Software
derivatives and A2ARs. The major ligand binding interactions are both polar and hydrophobic in nature and occur with residues in trans-membrane domains 3, 5, 6, and 7. Residues from the second extracellular loop (ECL2) outline the upper part of the binding cavity (Figure 2B). In our study, 7-(3-(4-(2fluoroethoxy)phenyl)propyl)-2-(furan-2-yl)-7H-pyrazolo[4,3e][1,2,4]triazolo [1,5-c]pyrimidin-5-amine 10a (FESCH, Scheme 2) and 7-(3-(4-(3-fluoropropoxy)phenyl)propyl)-2(furan-2-yl)-7H-pyrazolo[4,3-e][1,2,4]triazolo[1,5-c]pyrimidin5-amine 10b (FPSCH, Scheme 2) had binding modes that were similar to the cocrystallized 4-(2-[7-amino-2-(2-furyl)[1,2,4]triazolo[2,3-a][1,3,5]triazin-5-yl-amino]ethyl)phenol 17 (ZM241385, fifth compound in Table 1) conformation (Figure 2A), including the important hydrogen bond interactions with
■
RESULTS Molecular Docking. Table 1 shows the GOLD fitness scores and important interactions for the docked ligands. A molecular docking study was performed to elucidate the intermolecular interactions between the 7-(3-(4methoxyphenyl)propyl)-2-(2-furyl)pyrazolo[4,3-e]-1,2,4triazolo[1,5-c]pyrimidine-5-amine 8 (SCH442416, Scheme 2) 6767
dx.doi.org/10.1021/jm500700y | J. Med. Chem. 2014, 57, 6765−6780
Journal of Medicinal Chemistry
Article
Figure 2. (A) Interaction between antagonist 17 and A2AR binding site residues. (B) Docked conformation of 10b with A2AR, viewed from the membrane and extracellular (ECL) sides, showing ECL2-folded into the binding cleft. TMs 1−7 are labeled for reference. (C and D) Close-up view showing the pose for (C) 10a and (D) 10b in the A2AR binding site residues. The residues and compounds are shown in stick models. The residues involved in ligand binding are labeled and represented as violet sticks, oxygen atoms are shown in red, nitrogen atoms are shown in navy blue, and hydrogen atoms are shown in gray. The compounds are represented as purple sticks, the fluorine atom is represented in cyan, and the rest of the indicated atoms are similar to A2AR residues. The dashed lines in black indicate hydrogen bonds.
the active site residue Asn253 and a π−π stacking interaction with Phe168 of A2AR. The exocyclic free amino group of the pyrimidine ring structure of the tricyclic core and the oxygen of the furan ring make strong H-bond interactions with Asn253 of the receptor. This finding is consistent with the results from site-directed mutagenesis studies of the A2AR, which suggest that this amino acid is critical for ligand binding.50 Moreover, the π−π stacking interaction between Phe168 and His250 stabilizes the binding pose of the compound within the active site. Compound 8 derivatives, including the A2AR-bound crystal structure, 17, are oriented perpendicular to the plane of the cell membrane, with their flexible hydrocarbon side chain located in the extracellular domain. An analysis of ligand-bound crystal
structure and literature evidence47,51 suggests that ECL2 helps in ligand binding at the A2AR. We explored the impact of structural variability at the terminal phenolic position of 8 in the GOLD docking scores. As reported previously,47,51 conformational flexibility was also noted in our experiment; the terminal phenolic side chain forms a polar interaction with a crystallographic water molecule at the extracellular matrix of A2AR. As expected, N6-cyclopentyl-N8-isopropyl-N8,9-dimethyl9H-purine-6,8-diamine 18 (LUF5608, sixth compound in Table 1), a high affinity A1R antagonist and negative-control, yielded a very low docking score due to a lack of hydrogen bond formation with the active site residues of the receptor, further authenticating the findings of the docking study and indirectly 6768
dx.doi.org/10.1021/jm500700y | J. Med. Chem. 2014, 57, 6765−6780
Journal of Medicinal Chemistry
Article
Scheme 1. Synthesis of Fluoroalkyltosylatesa
Reagents and conditions: (i) tosyl chloride, pyridine, DCM, room temperature (RT), stirring 3 h; (ii) tetrabutyl ammonium fluoride trihydrate, acetonitrile, overnight reflux.
a
Scheme 2. Disconnection Retrosynthetic Scheme for the Reference Fluoroalkylated Compound 8 Analogs (10a and 10b)a
a
Reagents and conditions: (iii) BBr3, CH2Cl2, RT, 2 h; (iv) 5, Cs2CO3, MeOH, reflux, 1 h; and (v) 7, Cs2CO3, MeOH, reflux, 16 h.
[18F]fluoroalkylation of the phenol precursor 9 (Scheme 3). Table 2 lists the decay-corrected radiochemical yields, specific radioactivities, calculated partition coefficient (clogP) and experimentally determined distribution coefficient (LogD7.4) values for [18F]-10a and [18F]-10b. For both tracers, the radiochemical purity was >98%, and the total synthesis time, including quality control, was 114 ± 5 min (n = 18). The identities of the tracers were confirmed by spiking with authentic cold compounds in reversed-phase HPLC (RPHPLC). Ligand Metabolism. In Silico Metabolite Analysis. The predicted sites of metabolism are highlighted for parent compound 10b in Scheme 4 (data not shown for 10a and 8). The possible metabolic routes can be ranked in the following order: C-hydroxylation > N-oxidation > O-dealkylation. Human Liver Microsomal Metabolite Analysis. Table 3 summarizes the modifications that were detected with an
confirming the specificity of 10b and 10a toward A2AR over A1R. Chemical and Radiochemical Synthesis. The demethylation of commercially available 8 using boron tribromide (BBr3) resulted in a quantitative yield of 4-(3-(5-amino-2(furan-2-yl)-7H-pyrazolo[4,3-e][1,2,4]triazolo[1,5-c]pyrimidin7-yl)propyl)phenol, 9 (precursor of compound 8, Scheme 2).47 A retrosynthetic approach was adopted for the synthesis of reference standards (10a and 10b, Scheme 2), which were prepared by reacting the phenol precursor (9) with the appropriate fluoroalkyltosylate (5 or 7 in Scheme 1, selective fluoroalkyaltion) in 35% and 25% yields, respectively. It was adopted for the synthesis of reference standards (10a and 10b, Scheme 2). The two radiolabeled analogs, [18F]-10a and [18F]10b, were synthesized by a two-step two-pot procedure starting with the corresponding [18F]fluoroalkyl synthon ([18F]-5 or [18F]-7) made from [18F]fluoride and the appropriate ditosylate precursor (6 or 11), followed by the selective 6769
dx.doi.org/10.1021/jm500700y | J. Med. Chem. 2014, 57, 6765−6780
Journal of Medicinal Chemistry
Article
Scheme 3. Radiosynthesis of [18F]-10b and [18F]-10aa
a Reagents and conditions: (vi) K[18F]F−K2.2.2−K2CO3, acetonitrile, 6, 1,2-ethanediol di-p-tosylate, 125 °C, 10 min; (vii) tetrabutyl ammonium hydroxide (40% aq.), acetonitrile, sealed conditions, 115 °C, 15 min.
Table 2. Radiosynthesis and Lipophilicity Data clogP radioligand 18
[ F]-10a [18F]-10b a
a
radiochemical yield and purity (%)
specific radioactivity (GBq /μmol)
Chemsketch 12.01
Chembiodraw ultra 12.0
LogD7.4b
7 ± 2 and ≥98 8 ± 2 and ≥98
22.5 ± 5 136 ± 13
2.98 ± 0.98 3.27 ± 0.98
3.18 3.41
3.16 ± 0.03 3.41 ± 0.11
Overall radiochemical yields based on starting wet [18F]fluoride and corrected for decay. bExperimental value.
In Vitro Ligand Stability Test. The in vitro stability of the two [18F]tracers ([18F]-10a and [18F]-10b, Scheme 3) in different solutions such as PBS, saline, rat plasma, and human plasma was determined at 37 °C. After 1 and 2 h of incubation, radio-TLC analysis showed that 95−97% of both tracers were still intact, except in the saline solution. In the saline solution, multiple spots were observed as detected by radio-TLC for both tracers, and the radioactivity corresponding to the intact tracers was only 85−90%. In Vitro Autoradiographic Experiments. Figure 4 shows autoradiographic images of frozen rat brain sections that were incubated with [18F]-10a and [18F]-10b. In control sections, a clear difference was noted between the receptor-rich striatum and the receptor-poor cerebellum. The mean striatum-tocerebellum ratios were 2.75 ± 0.12 ([18F]-10a) and 2.99 ± 0.16 ([18F]-10b). In the presence of an excess (2 μM) of the A2ARspecific antagonist 8-[(1E)-2-(2-(3,4-dimethoxyphenyl)ethenyl]-1,3-diethyl-3,7-dihydro-7-methyl-1H-purine-2,6-dione, 12 (KW6002 in Figure 1), the binding of the tracer to the striatum was strongly reduced and the striatum-to-cerebellum ratio decreased to unity (n = 3). Specific binding, as assessed by a blocking study, was 57−62% and 64−67% of the total uptake in the striatum of [18F]-10a and [18F]-10b, respectively. Micro PET Images. PET images acquired after injection of [18F]-10a and [18F]-10b are presented in Figure 5. The two radioligands displayed similar regional distributions that corresponded to the known regional A2AR densities in the rat brain.7−9,11 In order to prove specific binding, we have used vehicle-control and blocker animals (Please refer in vivo and in vitro selectivity of the experimental section for more details). In
Scheme 4. Sites of Metabolism Predicted for 10b by SMARTCyp Webservicea
a
The same predicted metabolic sites are also applicable to compounds 8 and 10a.
ultrahigh-performance liquid chromatography/quadrupoletime-of-flight-mass spectrometer (UHPLC/Q-ToF-MS) and analyzed by Metabolynx after incubating compound 8 with human liver microsomes. The relative production of the different metabolites over time is presented in Figure 3. None of these modifications and major demethylated metabolites were detected in negative control and positive control (verapamil) incubations, respectively, with human liver microsomes. Under these conditions, the m/z ratio increased by 16, which is most likely the result of N-oxidation or Chydroxylation. The m/z ratio also decreased by 14, which may be due to a demethylation reaction. Moreover, a fluorinecontaining metabolite of 10b and free fluoride was observed, indicating defluorination, similar to previously reported results for 10a.47 6770
dx.doi.org/10.1021/jm500700y | J. Med. Chem. 2014, 57, 6765−6780
Journal of Medicinal Chemistry
Article
Table 3. Human Liver Microsomal Metabolite Analysis
cerebral uptake of the tracers was strongly reduced, and regional differences in tracer uptake were no longer observed. In Vivo Radioligand Kinetics and Metabolism. Kinetics of Radioactivity in Brain. The cerebral kinetics of radioactivity after the injection of [18F]-10a and [18F]-10b are presented in Figure 6 (panels A−D). In vehicle-treated control animals (n = 6), the uptake of radioactivity rapidly increased to a maximum (2.5 min after injection), which was followed by an exponential washout (panels A and C). In animals treated with 12 (n = 6), the cerebral uptake of 18F was strongly reduced, and the radioactivity was rapidly washed out from all brain regions (panels B and D). The difference in the striatal uptake of 18F in control and pretreated rats was statistically significant at most time points. We estimate receptor occupancy on the basis of PETstandardized uptake values (PET-SUVs) (at the time of maximum uptake), reported A2AR densities (953 fmol/mg protein in rat striatum52) and injected tracer doses in nanomolar amounts. Assuming that brain tissue contains 10% protein, we calculated that less than 2% and 8% of the cerebral A2AR population was occupied by [18F]-10b and [18F]-10a, respectively, in both control- and blocker-treated rats. Radioactivity Kinetics in Plasma. A rapid, biexponential plasma clearance was observed in all groups. Pretreatment did
Figure 3. Relative amounts of several modifications over 90 min during incubation of (8) with human liver microsomes. The error bars indicate the standard deviation.
vehicle-control animals, the striatum was clearly visualized. The extrastriatal binding of both tracers was hardly visible, but strong uptake was observed in the skull bone. When animals were pretreated with the A2AR antagonist 12 (1 mg/kg), the 6771
dx.doi.org/10.1021/jm500700y | J. Med. Chem. 2014, 57, 6765−6780
Journal of Medicinal Chemistry
Article
Figure 4. Autoradiographic images of sagittal sections of rat brains after 90 min of incubation with (A and B) [18F]-10b or (C and D) [18F]-10a in the (B and D) presence or (A and C) absence of an excess of a known A2AR-selective antagonist, 12 (2 μM).
■
DISCUSSION We have evaluated [18F]-10a and [18F]-10b as PET tracers for the cerebral imaging of A2ARs; these tracers may provide many logistic advantages and can be used in centers without an onsite cyclotron. Thus, we synthesized fluorinated molecules based on a pyrazolo-triazolo-pyrimidine template (compound 8) that is known to cross the BBB because of its appropriate lipophilicity (clogP = 2.9), molecular weight, charge, and hydrogen bonding.39,43,44 The molecular docking approach provides valuable atomic-level insight into the behavior of a small molecule in the binding site of the protein. It also provided insight into the binding mode of compound 8 derivatives to the active site of the receptor. Compound 8 and its fluoro analogs (10a and 10b) had better GOLD fitness scores than the clinically studied PET tracer (E)-8-(3,4,5trimethoxystyryl)-1,3-dimethyl-7-[11C]methylxanthine ([11C]KF18446) and the A2AR-bound crystal structure 17. GOLD scores are good indicators to predict the binding orientations of compound 8 derivatives. A higher score predicts better binding orientations with the receptor residues. We have used the A1R decoy, 18-as negative control, to validate the quality of the GOLD scoring function.53 Small structural changes of the phenoxy substituent appeared to be well-tolerated, and this is substantiated by our in vivo study. Subtype selectivity can be expected because the lead compound 8 shows a >10,000-fold selectivity for the A2AR subtype compared to other AR subtypes. Moreover, the binding affinities of 8 (Ki = 0.5 nM) and 10a (Ki = 12.4 nM) to the A2AR were adequate for imaging.43,47 Even though the reported Ki value of 53.6 nM for 10b54 is much lower than that of the lead compound 8, the affinity data predict faster clearance than 8 and 10a and more preferable brain kinetics for the quantitative evaluation of the
not significantly affect the clearance of radioactivity from the plasma compartment. In Vivo Metabolite Analysis. An unidentified radiometabolite with a Rf value of 0−0.1 was observed in rat plasma (the Rf value of authentic [18F]-10b was 0.6). The fraction of total plasma radioactivity representing the parent compound decreased to 66 ± 16% at 60 min and 53 ± 20% at 90 min. Pretreatment with 12 did not affect the rate of tracer metabolism. The fraction of total plasma radioactivity representing [18F]-10a was 46 ± 17% at 60 min and 36 ± 14% at 90 min. Ex Vivo Biodistribution Data. The biodistribution data for both tracers are shown in Figure 7. After pretreatment with 12, the uptake of both compounds was reduced in the A2AR-rich striatum (approximately 69% of [18F]-10b and 45% for [18F]10a). For [18F]-10b, the effect of the blocker was statistically significant in the frontal cortex and striatum, whereas for [18F]10a (n = 3), no significant effect of the blocker was observed in any of the brain regions; however, the greatest decrease of the tracer uptake was observed in the striatum. Striatum-tocerebellum ratios can be used as indices for the in vivo binding of the tracers to A2ARs. Striatum-to-cerebellum ratios of 3.5 and 2.1 were reached at 106 min postinjection for [18F]-10b (n = 6) and [18F]-10a (n = 3), respectively. The ratios of the uptake in other regions to the cerebellum were approximately equal to one. The standard uptake values in the skull bone (2.06 ± 0.58) for [18F]-10b were significantly higher than those of [18F]-10a (0.29 ± 0.05). For both tracers, the plasma-to-blood ratio was greater than one, and the negligible binding to red blood cells in pretreated and control animals indicated that the radioligands preferentially distributed to the plasma. 6772
dx.doi.org/10.1021/jm500700y | J. Med. Chem. 2014, 57, 6765−6780
Journal of Medicinal Chemistry
Article
Figure 5. Small-animal PET images of a coronal plane of rat brains after injections of (A and B) [18F]-10a or (C and D) [18F]-10b. The images represent the summed frames from 17 to 90 min post injection. (A) Vehicle-control (left); (B) a compound 12-treated animal ([18F]-10a) (right). (C) Vehicle-control (left); (D) a compound 12-treated animal ([18F]-10b) (right). The images were normalized for body weight and injected dose.
ligand−receptor binding. The pyrazolo-triazolo-pyrimidine scaffold allows for the easy and quick incorporation of an [18F] label in the acidic phenol group, and this phenoxy substituent can also be used to modify the lipophilicity of the compound. Fluoroalkyl chain lengthening beyond the fluoropropyl substituent results in a higher molecular weight (MW) and lipophilicity for a compound. It has been suggested that the MW should be kept below 450 Da to facilitate brain penetration with fewer side effects such as high rapid metabolic turnover, poor absorption, and toxicity.55 High lipophilicity causes unacceptable binding to plasma proteins, decreasing the free drug concentration available to pass the BBB, or binding to hydrophobic protein targets other than the desired one, resulting in high levels of nonspecific binding in the brain.39,55 On the basis of these considerations, 10b was selected as a novel candidate for radiolabeling to obtain the expected lipophilicity and a MW that ensures the crossing of the BBB. Computational prediction of the sites of cytochrome P450 (CYP450)-mediated metabolism and an in vivo plasma radioTLC metabolite analysis indicated the formation of polar metabolites, which are not expected to cross the BBB. The results obtained after incubation of 10b with hepatic microsomes were in good agreement with a previously reported
experiment and the predictions of a two-dimensional (2D) method (SMARTCyp) describing CYP450-mediated drug metabolism.47,56 In contrast to the results obtained from a SMARTCyp prediction of drug metabolism (please refer to Ligand Metabolism module in the Results section), the metabolic routes of compound 8 after incubation with liver microsomes can be ranked in the following order: Odealkylation > parent-C10H12O > N-oxidation or C-hydroxylation > deamination (Figure 3, Table 3). A stability test indicated that both 10a and 10b are highly stable in vitro. Because of the observed multiple spots as detected with radioTLC in a saline solution, we used PBS instead of saline in the formulation of the tracers. The fraction of total plasma radioactivity representing [18F]-10b was 18−20% higher than that of [18F]-10a at both 60 and 90 min in our in vivo metabolite analysis. However, stronger skull bone radioactivity uptake (i.e., stronger defluorination) was observed with [18F]10b than [18F]-10a in both an ex vivo biodistribution study and a small-animal PET (microPET) image analysis. In our imaging studies (especially with [18F]-10b), the accurate quantitation of the radioactivity in the frontal cortex was difficult due to spillover from radiofluorine in the skull bone. A retrosynthetic approach for 10a and 10b synthesis was successfully applied to avoid a cumbersome and time6773
dx.doi.org/10.1021/jm500700y | J. Med. Chem. 2014, 57, 6765−6780
Journal of Medicinal Chemistry
Article
Figure 6. (A−D) Kinetics of (A and B) [18F]-10a and (C and D) [18F]-10b-derived radioactivity in the rat brain. The error bars indicate the SEM. (A and C) Vehicle-control animals (left); (B and D) compound 12-treated animals (right). ●, striatum; ▼, cerebellum.
Our radiosynthetic procedure for [18F]-10a is much faster (by 15−20 min) than the existing procedure.47 The applied radiolabeling approach is versatile; we can quickly adopt the same procedure for the radiosynthesis of both compounds [18F]-10a and [18F]-10b. The radiochemical purities were also adequate and amounted to more than 98% of the total radioactivity as determined by UHPLC quality control. In vitro autoradiography (ARG) confirmed the selectivity of [18F]-10a and [18F]-10b for A2A Rs. The tracer binding pattern, especially in the striatum and other parts of the brain, was comparable with ex vivo biodistribution readings. The regional distribution of radioactivity in the rat brain after the injection of [18F]-10a and [18F]-10b also suggests that these tracers are capable of measuring regional A2AR densities. After pretreatment with a subtype-selective xanthine antagonist, 12, the tracer uptake in the striatum was greatly suppressed, and regional differences were no longer present. In the biodistribution and PET studies, the uptake of [18F]-10b in the cerebellum and frontal cortex (areas lacking A2ARs) was also decreased after pretreatment with 12. The most logical explanation for the specific binding in the cerebellum is that
consuming scheme involving 8 reaction steps. In the synthesis of compound 10b, 3-fluoropropyl tosylate 7 yielded a slightly better result than 1-bromo-3-fluoropropane (25% vs 19% yield) because tosylate is a better leaving group than bromide. Tracers were successfully synthesized using a two-pot two-step procedure (Scheme 3). The [18F]fluoroalkylation of phenol precursor 9 using corresponding intermediate fluorosynthons ([18F]-5 or [18F]-7) yielded the desired ligands [18F]-10a and [18F]-10b at moderate yields (7−8%) and satisfactory specific activities (Table 2). The average radiochemical yield of the [18F]-5 or [18F]-7 obtained was 50 ± 5%, whereas the final fluoroalkylation conversion was approximately 25 ± 5%. Purification by HPLC and solid phase extraction provided a decay-corrected radiochemical yield of 7−8%. The long evaporation step of the captured eluate [14 mL of hexane/ ether (3:1)] during the purification of the [18F]-5 or [18F]-7 and the losses that occurred during the other manipulations of the synthesis accounted for the moderate radiochemical yields of the synthesized tracers [18F]-10a and [18F]-10b. The purification of the [18F]-5 or [18F]-7 by RP-HPLC and C-18 light Sep-Pak columns may improve the radiochemical yield. 6774
dx.doi.org/10.1021/jm500700y | J. Med. Chem. 2014, 57, 6765−6780
Journal of Medicinal Chemistry
Article
Figure 7. Cerebral biodistribution data of [18F]-10b and [18F]-10a at 106 min after injection. The error bars indicate the SEM. BOLF = Bulbus olfactorius, CERE = Cerebellum, FCor = Frontal cortex, Stria = Striatum, Hipp = Hippocampus, Medu = Medulla, PTOC = Parietal/Temporal/ Occipital Cortex.
Figure 8. Striatum-to-cerebellum ratios of [18F]-10b and [18F]-10a as a function of time. The solid and broken lines represent [18F]-10b and [18F]10a, respectively. The error bars indicate the SEM.
pharmacological effects or change in the physiological states of the A2AR system during our microPET study. The striatal uptake of both [18F]-ligands was clearly visualized using PET scans; both tracers reached a striatumto-cerebellum ratio of approximately 4.6, which is similar to the experiments with [11C]-8 result (4.6 ± 0.27).44 However, the maximum ratio for [18F]-10b was reached at a later time point (37 min) than that of [18F]-10a (25 min) and [11C]-8 (15 min), most likely because of the higher lipophilicity of [18F]10b (Figure 8). Lipophilicity may prolong the circulating halflife of a tracer, resulting in extended availability for binding to A2ARs. After the maximum had been reached, the concentration of [18F]-10a in the brain remained fairly stable until 30 min after injection (similar to [11C]-8), whereas the concentration
the endothelium and blood vessels express A2ARs, even if the brain tissue does not.18,57 Taking into account the findings that pretreatment with 12 reduces the distribution volume of candidate reference tissues such as the cerebellum, we choose not to quantify the blocking effect using a (simplified) reference tissue model, 2-tissue compartment model, and Logan analysis. It is evident from the microPET images that the extrastriatal binding of the [18F]-tracers was barely visible, except for the strong uptake in skull bone due to defluorination, whereas in the case of xanthine PET ligands, the extrastriatal retention of radioactivity was visualized in the cortex, cerebellum, and thalamus.34 Because of estimated receptor binding of the injected tracer mass (
